• DocumentCode
    1784843
  • Title

    Tessellation-based coarse registration method for 3D reconstruction of the female torso

  • Author

    Costa, Pyramo ; Monteiro, Joao P. ; Zolfagharnasab, Hooshiar ; Oliveira, Helder P.

  • Author_Institution
    Fac. de Eng., Univ. do Porto, Porto, Portugal
  • fYear
    2014
  • fDate
    2-5 Nov. 2014
  • Firstpage
    301
  • Lastpage
    306
  • Abstract
    The medical procedures related with the Breast Cancer Conservative Treatment (BCCT) have evolved towards the usage of affordable and practical tools, along with the recent inclusion of volumetric information of the breast. A richer three-dimensional (3D) model of the female torso allows, for instance, improvement of the evaluation the aesthetic outcome of BCCT and the surgery planning. The standard 3D reconstruction methods often fail to model objects of interest using highly misaligned views. In this work, a Tessellation-based coarse registration method is proposed, based on robust keypoints extraction from RGB-D data using the Delaunay Triangulation (DT) principle. With this method, it is possible to reconstruct female torso data with detail using only 3 views, in feasible time. Structures such as the nipples and the breast contour were correctly reconstructed and a highly correlated with reference models.
  • Keywords
    cancer; image colour analysis; image registration; medical image processing; mesh generation; 3D female torso model; 3D reconstruction methods; BCCT; Delaunay triangulation principle; RGB-D data; Tessellation-based coarse registration method; breast cancer conservative treatment; breast contour; keypoints extraction; surgery planning; Breast; Histograms; Image color analysis; Iterative closest point algorithm; Standards; Three-dimensional displays; Torso;
  • fLanguage
    English
  • Publisher
    ieee
  • Conference_Titel
    Bioinformatics and Biomedicine (BIBM), 2014 IEEE International Conference on
  • Conference_Location
    Belfast
  • Type

    conf

  • DOI
    10.1109/BIBM.2014.6999173
  • Filename
    6999173